1. Field of the Invention
The present invention relates to hermetically sealed semiconductor chips. More particularly, the present invention relates to coating at least the active surface of a flip chip with a silicon nitride sealing layer.
2. State of the Art
Chip On Board (xe2x80x9cCOBxe2x80x9d) techniques are used to attach semiconductor dice to a printed circuit board, including flip chip attachment, wirebonding, and tape automated bonding (xe2x80x9cTABxe2x80x9d). A flip chip is a semiconductor chip or die that has a pattern or array of spaced terminals or pads on its active surface for face-down mounting of the chip to a printed circuit board or other conductor-carrying substrate. Generally, the flip chip active surface carries one of the following types of electrical connector elements: Ball Grid Array (xe2x80x9cBGAxe2x80x9d)xe2x80x94wherein an array of minute solder balls (sometimes called C4 connections, for controlled-collapse-chip-connect) or other conductive material is disposed on the surface of a flip chip that attaches to the substrate trace terminals or connecting pads (xe2x80x9cthe attachment surfacexe2x80x9d); or a Slightly Larger than Integrated Circuit Carrier (xe2x80x9cSLICCxe2x80x9d)xe2x80x94which is similar to a BGA, but having a smaller solder ball/conductive material pitch and diameter than a BGA. With the BGA or SLICC, the solder or other conductive ball or element arrangement on the flip chip must be a mirror image of the connecting pads on the printed circuit board so that precise connection is made. When solder balls are employed, the flip chip is bonded (electrically and mechanically connected) to the printed circuit board by reflowing the solder balls. Other conductive elements such as conductive epoxies or conductor-filled epoxies or other polymers may be employed in lieu of solder balls and heat-cured after chip attachment.
Semiconductor chips must be able to withstand a wide variety of environmental conditions such as moisture, ion bombardment, heat and abrasion. A significant amount of work has been directed toward various protective measures to minimize the exposure of semiconductor chips to these environmental conditions in order to increase their reliability and operating life.
Many prior art processes for protecting semiconductor chips have involved sealing or encapsulating the chips after they have been attached to their respective lead frame, printed circuit board, or other carrier substrate. Plastic encapsulation of semiconductor chips is currently the most common form of packaging chips. Plastic encapsulation normally consists of encasing a leadframe-mounted semiconductor die in plastic under pressure in a transfer molding process. Furthermore, so-called xe2x80x9cglob topxe2x80x9d (commonly silicones and epoxies) and underfill (commonly epoxies) materials have been used to protect chips secured on a printed circuit board (such as an FR-4 glass-epoxy board), or ceramic or silicon substrate. A non-conductive polymer underfill is generally disposed between the active surface of a xe2x80x9cflippedxe2x80x9d semiconductor chip and the printed circuit board or other carrier substrate for environmental protection and to enhance the mechanical attachment of the semiconductor die to the substrate. An overfill encapsulant of viscous liquid or gelled silicone or epoxy (glob top) is sometimes applied over an entire assembly after COB attachment. In short, it is known in the art to use layers of silicones, polyimides, epoxies, plastics, and the like for protection of the COB assemblies.
While transfer-molded plastic encapsulation and glob tops are effective methods of protecting the semiconductor die from abrasion and some mechanical damage, such approaches are of limited value, since most such packaging structures are permeable to environmental moisture and ions to a measurable degree. This permeability, however slight, leaves the semiconductor chip susceptible to degradation from electrochemical reactions with atmospheric contaminants. The numerous and extensive polymer/metal interfaces at the lead entries of an encapsulated semiconductor package afford ample opportunities for moisture ingress as well as allowing soluble ions present to provide an electrolyte for a corrosive failure mechanism of the semiconductor chip. Also, the extensive use of precious metals coupled with base metals in chips and packages provide DC galvanic potentials for electrochemical corrosion reactions and dendrite growth, which affect the performance and life of the encapsulated semiconductor chip.
As a result of the problems associated with the plastic encapsulation of semiconductor chips, it has been established as desirable to hermetically package chips to prevent external moisture and chemicals from contacting the semiconductor chip. U.S. Pat. No. 5,136,364 issued Aug. 4, 1992 to Byrne relates to hermetically sealing semiconductor chip bond pads to prevent moisture contamination along the interface of the multiple metal layers of typical bond pads. The hermetic sealing comprises layers of passivation materials wherein the upper passivation layer is a silicon dioxide and nitride combination or silicon carbide. The hermetic sealing can also include layers of glass and/or polyimide.
It is known to seal semiconductor chip active surface circuitry at the wafer stage of production by applying a passivation coating to the wafer with ceramic materials such as silica and/or silicon nitride by chemical vapor deposition (xe2x80x9cCVDxe2x80x9d) techniques. See, for example, U.S. Pat. Nos. 5,046,161, 5,084,752 and 5,406,122. However, the subsequent etching back of the passivation coating at the bond pads of the semiconductor chip may damage the passivation coating adjacent the bond pads, thereby affecting the reliability of the chip and shortening the life of the chip due to environmental corrosion. The sides of the passivation coating etched from the bond pad edges can also be permeable.
In an attempt to hermetically seal semiconductor chips without the use of external packages, U.S. Pat. No. 5,481,135 issued Jan. 2, 1996 to Chandra et al. suggests the use of lightweight ceramic protective coatings, such as those derived from hydrogen silsesquizane and silicate esters. These coatings are applied to the active surface of a semiconductor chip at a wafer level. Although the bond pads are subsequently exposed by removing a portion of the ceramic protective coating, the resultant circuits are purported to remain hermetically sealed. However, the process of this patent requires the application of a diffusion barrier metal layer to protect the bond pads during the etching to expose the bond pads.
A room temperature plasma deposition system capable of applying a low-stress silicon nitride over components on a circuit assembly is disclosed by L. Gates in xe2x80x9cSealed Chip-on-Board,xe2x80x9d Electronic Packaging and Production, September 1994, pp. 48-50 (the xe2x80x9cGates articlexe2x80x9d). In the described deposition process, a semiconductor die which is wire bonded to a substrate is subsequently entirely coated with silicon nitride. Thus, the silicon nitride covers the bond pads, bond wires, and other components of the assembly.
The disclosures of hermetically sealed semiconductor chips described above, with the exception of the Gates article, fail to provide a process for sealing a wafer or semiconductor chip without damage to the semiconductor chip or bond pads from back etching to expose the bond pads, unless additional processing steps are employed. Gates, moreover, does not address the complexities of hermetically sealing a flip chip type semiconductor die assembly. Therefore, it would be advantageous to develop a technique for simply, quickly and inexpensively forming a hermetic seal on in combination with commercially-available, widely-practiced semiconductor device fabrication techniques compatible with flip chip attachment.
The present invention relates to an apparatus and a method for providing a silicon nitride (Si3N4) hermetic seal over at least a portion of a flip chip prior to attachment to a substrate. The present invention also contemplates and includes sealing the flip chip and substrate with a silicon nitride layer after the attachment of the flip chip to the substrate. The present invention is achieved by providing a semiconductor chip having a plurality of bond pads on an active surface, wherein at least some of the bond pads carry a solder ball or other conductive material (hereinafter xe2x80x9cconductive connectorxe2x80x9d) disposed thereon (e.g., BGA or SLICC type arrangements).
A layer of silicon nitride, preferably a low temperature silicon nitride, is deposited on the active surface of the semiconductor chip to cover the exposed semiconductor die active surface and conductive connectors surmounting the bond pads thereon. The silicon nitride layer is preferably applied at a temperature of about ambient or room temperature to about 200xc2x0 C., depending on the material of the printed circuit board or other carrier substrate to a thickness of about 1 to 2 xcexcm by CVD. These parameters usually require 600 seconds or less to complete the deposition. Optionally, the silicon nitride layer can be applied by plasma enhanced CVD deposition or other deposition techniques known in the art. The room-temperature deposition process of the Gates article is suitable, particularly for FR-4 and other temperature-susceptible boards. The silicon nitride layer, without additional materials or processing, provides protection from moisture-driven ionic corrosion. The active surface of the semiconductor chip being sealed, it is thus protected from contamination and can be stored until needed in the semiconductor device fabrication process.
It is preferable to have the silicon nitride layer extend over the peripheral edges of the active surface of the semiconductor chip. By extending the silicon nitride layer over the semiconductor chip peripheral edges, the otherwise-exposed interfaces between the layers of circuitry carried on the semiconductor chip active surface and the interface between the circuitry and the underlying silicon (or other semiconductor substrate material, such as gallium arsenide) on which the circuitry is printed are hermetically sealed, this approach virtually eliminating the possibility of moisture or other contaminants infiltrating.
It is preferable to clean the semiconductor chip prior to coating with the silicon nitride layer. The semiconductor chip is preferably plasma cleaned with an oxygen (ozone) plasma system which removes organics and other contaminants from the surface of the semiconductor chip by pulling carbon into CO2. As an alternative, solvent or other cleaning methods may be employed.
Before flip chip attachment to the carrier substrate, a portion of the silicon nitride layer covering each conductive connector is removed to expose a portion of the conductive connector material. The removal of the silicon nitride portion can be achieved through etching, such as by dipping the outer tips of the conductive connectors in an etch solution. A preferred etching solution is a phosphoric acid solution at a temperature of about 100xc2x0 C. to about 140xc2x0 C.
The removal of the silicon nitride portion at the tips or free ends of the conductive connectors can also be achieved through mechanical abrasion, optionally assisted by chemical reaction. A preferred abrasion technique is the use of a rotating polishing pad, as employed in a technique known in the industry as chemical mechanical polishing (planarizing) or xe2x80x9cCMPxe2x80x9d (see U.S. Pat. 5,449,314 issued Sep. 12, 1995 to Meikle et al.). CMP involves holding a semiconductor chip against a rotating wetted polishing platen under controlled chemical, pressure and temperature conditions. Typically, an aqueous colloidal silica solution is used as the abrasive fluid. The polishing or workpiece material removal mechanism is a combination of mechanical action and chemical reaction, the surface of the workpiece being polished with the aqueous solution. Employing the CMP technique in the present invention also provides the additional benefit of achieving a relatively precise and uniform connector height for all connectors extending from the active surface of the wafer being processed.
The removal of the silicon nitride portion can also be achieved by fracturing the silicon nitride at the conductive connector tips by transverse impact against a hard surface. To effect the required fracture, the silicon nitrided connector tip portions of a wafer are simply impacted against the surface (xe2x80x9cspankedxe2x80x9d) with sufficient force to fracture the silicon nitride on the desired end portions of the conductive connectors. The spanking may either simply crack the silicon nitride, so that during solder reflow or epoxy cure, the conductive connector material penetrates the cracks to bond to the substrate terminal packs, or may completely break the silicon nitride off the end portions of the conductive connectors. The target impact surface may also be vibrated with ultrasonics for a more efficient xe2x80x9cscrubbingxe2x80x9d type removal of the silicon nitride portion during the spanking process.
As implied, it is, of course, understood that the above process steps can be, and desirably are, formed on a wafer level. The semiconductor chips may be diced immediately after the application of the silicon nitride layer or after the subsequent re-exposure of the conductive connector portion, although the latter approach is preferable. When a wafer is coated prior to dicing, it is preferable to scribe or etch a bevelled channel or trough around the boundary of each semiconductor die location on the wafer prior to coating. The bevel extends through the circuit layers on the active surface into the underlying substrate. Thus, when the silicon nitride layer is deposited over the wafer active surface, the layer extends into the bevels (channels) and partially covers what will be the semiconductor die active surface peripheral edges when the wafer is subsequently diced. With this technique, the interfaces between the layers of circuitry on the semiconductor die active surface and between the circuitry and the underlying silicon on which the circuitry is formed are sealed, virtually eliminating the possibility of contaminant infiltration.
A semiconductor device according to the invention is assembled by attaching the flip chip to a substrate having a plurality of terminal or connection pads on an active surface arrayed in a mirror image of the bond pads/conductive connectors of the silicon nitride coated semiconductor chip. The carrier substrate may comprise ceramic, silicon, polyimide or other polymers, FR-4, or similar materials, the particular substrate materials being insignificant to practice of the invention except as to temperature and time limitations affecting the nitriding process employed. Subsequent to flip chip attachment, the entire assembly is silicon nitrided, again preferably after plasma cleaning. It is noted that the silicon nitride CVD of the present invention is preferably performed at a temperature of about 50xc2x0 C. to about 200xc2x0 C., such process being well known in the art. While FR-4 (glass-epoxy laminate) board material has an upper temperature limit of about 175xc2x0 C. before degradation is experienced, it has, however, been found by the inventors that the FR-4 material is sufficiently robust to withstand temperatures between 185xc2x0 C. and 190xc2x0 C. for short periods of time sufficient to achieve acceptable CVD of the silicon nitride. It is also noted that the use of a polyimide substrate material in combination with a silicon nitride coating as set forth in the present invention forms a hermetic seal with equivalent performance characteristics to a preformed ceramic package.
The nitrided flip chip is attached to the substrate by contacting the exposed conductive connector portions with their respective substrate bond pads. With the use of solder balls, the solder is reflowed to attach to the substrate terminal pads. With epoxy conductive elements, a heat cure is effected. The reflowed solder balls or cured epoxy elements are protected from corrosion by the silicon nitride layer without the need for an underfill encapsulant. Once the flip chip is attached to the substrate for further protection and sealing, a second passivation layer may be applied to the entire substrate/flip chip assembly. The second passivation layer is also preferably silicon nitride applied in the manner described above.
A secondary advantage of the use of silicon nitride is that the relatively small thickness required for it to effectively function as an encapsulant results in low stress in the film. Thus, the application of the silicon nitride layer will not cause warpage of the substrate that may occur in some instances with the use of other forms of semiconductor chip encapsulation, such as the previously mentioned xe2x80x9cglob top,xe2x80x9d in combination with a carrier substrate of marginal structural rigidity.
It is, of course, understood that the semiconductor chip need not be coated (nitrided) prior to attachment to the substrate. The semiconductor chip can be first flip chip attached with the silicon nitride layer applied thereafter to the entire assembly.